Data
Download MUSCLE-VCD-2007
The corpus used for CIVR 2007 video copy detection evaluation is referred as MUSCLE-VCD-2007. It is freely available on demand with the following terms of use. Please contact us.
The terms of use of the MUSCLE/CIVR video copy detection evaluation corpus 2007 (MUSCLE-VCD-2007) are the following:
-The corpus is for research purposes only.
-The origin of the corpus must be precised in any publication related to its use: MUSCLE-VCD-2007, http://www-rocq.inria.fr/imedia/civr-bench/index.html.
-Any publication related to the use of the corpus has to be referenced on this website.
-The videos themselves come from two websites www.archive.org and www.open-video.org The terms of use can be found here: archive.org terms of use, open-video.org terms of use.
Organization
The video database was available for all the teams two months before the event.
On the day of the event, the participants bring a computer with their video copy detection system and all needed data. Internet access is not available.
The queries are given on a DVD which contains a directory with all the query video files. The "searchers" designated by each team use their system to find the copies, there is no pre-processing of these videos before the "D" day.
When the system finishes to process the videos, the results are given to the organization team on USB key. The time used for dealing with all the videos on a sub-task is measured between the moment the DVD was given to the participant and the moment the USB Key is given back to the organizer.
Video Database
About 100 hours of video materials comming from different sources: web video clips, TV archives, movies. The videos cover very large kind of programs: documentaries, movies, sports events, TV shows, cartoons etc. The videos have different bitrates, different resolutions and different video format. These videos have been provided in their original format and also in an MPEG1 format by a reencoding.more details
Video Queries ST1
- 15 videos with transformations
- total length of queries: 2 hours 30 mins
Metric :
a set of video is used as queries and each query returned an answer: the file is a copy of a video (or of a part of a video) in the database or the file was not a copy. The result files should contain as many lines as the number of video queries. Each line should be as follows:
| <Query> | <Ref> |
- <Query> is the name of the query video file
- <Ref> is the name of the corresponding reference video file in the database or "not_in_db" if no copy is found
Quality = Ncorrect / Nqueries
groundtruthST1
Video Queries ST2
Transformed extract have been inserted in videos not in the database. The goal is to find these segments with the boundaries despite the transformations.- 3 query videos
- total length : 45 mins
Metric:
The result files for comparing the results of each system contain as many lines as detected segments with every line as follows:
| <Query> | TcIn | TcOut | <Ref> | TcRef |
- <Query> is the name of the query video file
- TcIn is the beginning time code of the detection in the query video
- TcOut is the end time code of the detection in the query video
- <Ref> is the name of the corresponding reference video file in the database
- TcRef is the beginning time code of the detection for the reference video
There are two final criteria:
- One is computed from the percentage of mismatched video segments in all queries.
- The other is computed from the percentage of mismatched frames in all queries (due to either non-detected segments, or imprecisions in the detection or false positives).
groundtruthST2